Key takeaways
- Track share of answers per AI tool, not one blended score.
- Run each prompt more than once; one run proves little.
- Record cited sources: they show where to earn mentions.
- Link AI visibility to GA4 referrals and CRM leads.
Why rankings are not enough
A buyer who asks ChatGPT for a shortlist never sees page one. If your brand is not in the answer, a good Google ranking does not help with that buyer. AI answers also change from run to run, so you need metrics built for answers that move.
The 7 KPIs, with formulas
1. Share of answers (mention rate)
Runs in which your brand is named, divided by all runs, per AI tool. Example: named in 14 of 20 runs on one tool = 70%.
2. Top-3 rate
Runs in which you are among the first three brands named, divided by all runs. Buyers rarely read further.
3. Share of voice
Your mentions divided by all brand mentions in the same runs, for you and the competitors you track.
4. Answer stability
Of the prompts where you are named at least once, the share where you are named in every run. Low stability means a weak position.
5. Cited sources
The domains each tool cites, counted across runs, and the share that are your own pages. Third-party sources you are missing from become your mention plan.
6. Brand accuracy
Answers that state your services, markets and locations correctly, divided by answers that describe you. Wrong facts lose deals quietly.
7. AI referral traffic and leads
Sessions from chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com in GA4, and the leads and opportunities they bring in your CRM.
